|
NEG Food Manufacturing Hub
|
| |
|
Issue #004 · Scale-Up Engineering
From Farmers Market to 50,000 Units: The Real Engineering of Scale-Up
A product that sells at a farmers market is not yet a business model. It is a proof of concept.
The real test begins when demand rises and the operation has to hold together under pressure. At 50,000 units, the business is no longer about charm, hustle, or founder instinct. It is about systems, tolerances, throughput, and control.
The first production run comes back wrong — texture off, colour different, shelf life not matching spec. Nothing in the recipe changed. Most scale-up failures aren’t R&D failures. They’re engineering translation failures. This issue explains why.
|
|
|
In This Issue
→ | Why small scale hides big problems — and what breaks first |
→ | The three physics principles that change when vessel size increases |
→ | Three real-world failure modes and the root cause behind each |
→ | The engineering toolkit: dimensionless analysis, pilot data, co-manufacturer briefing |
→ | The four questions that determine whether you’re actually ready to scale |
|
|
“Scale does not create new problems. It exposes the ones that were already there.”
|
|
The Business Reality
Small Scale Hides Big Problems
At market scale, you can get away with a lot. You adjust batches by eye. You fix defects manually. You explain inconsistencies face to face. But once volume increases, every weakness becomes visible. A slightly unstable formulation becomes a spoilage problem. A weak package becomes a returns problem. A loose process becomes a cost problem.
This is where many teams get caught. They think scaling means making the same thing, just faster. It does not. Scaling means redesigning the product and the operation so they can survive higher volume without collapse. The question changes from “Can we make it?” to “Can we make it the same way, every time, at higher speed, with less waste?”
|
|
What Actually Breaks First
01 | The recipe is not robust enough for commercial equipment |
|
02 | The packaging cannot protect the product in transit or on shelf |
|
03 | The line is too dependent on skilled labour or founder intervention |
|
04 | Quality control only happens after the problem has already spread |
|
05 | Inventory and raw material planning are too loose for larger orders |
|
|
|
Scale-up is not just an operations issue. It is a system design issue. Heroics do not scale. Systems do.
|
|
|
|
The Engineering Explanation
The Three Physics Principles That Change at Scale
A 10-litre batch in a domestic pan behaves very differently from 1,000 litres in a jacketed vessel. Same recipe. Same ingredients. Completely different physical environment. Three principles are responsible for most of what goes wrong.
|
01 — Surface Area to Volume Ratio (SA:V) |
|
Your artisan pan at 10L had a high ratio of heated surface to product volume — heat reached the product quickly and evenly. A 1,000L jacketed vessel has a fraction of that ratio. The result: longer heat-up times, slower cool-down, extended hold periods at temperature.
Everything that depends on a thermal profile — texture, colour, pectin activation, microbial kill — shifts accordingly. Your CCP critical limits must be revalidated against the new vessel geometry.
|
|
02 — Mixing Dynamics and Flow Regime |
|
“Mix on medium speed” is not a transferable instruction. What matters is the flow regime inside the vessel — characterised by the Reynolds number. The same impeller RPM that gave gentle mixing in a 50L pilot vessel can create destructive shear or dead zones in a 2,000L production tank.
For emulsified products this is critical — a shear profile that stabilises an emulsion at one scale can destroy it at another. Ask your co-manufacturer for impeller geometry, tip speed, and power draw data — not just vessel volume.
Re = ρND²/μ |
Reynolds Number — matches flow regime across scales |
P/V = const. |
Power per unit volume — most common scale-up criterion (0.1–2 kW/m³) |
Nu = hD/k |
Nusselt Number — predicts how jacket efficiency changes with geometry |
|
|
03 — Thermal Mass and Process Inertia |
|
Large vessels resist temperature change. A 2,000L vessel has significant thermal inertia — it takes time to heat, it overshoots setpoint, and it takes time to cool. Heat treatment validation (F&sub0; or pasteurisation units) must account for actual come-up time in the production vessel — not a pilot or spec sheet calculation.
Practical implication: Log temperature at the geometric centre of the product mass, not just the jacket outlet or wall thermocouple. That is your real process data.
|
|
|
|
Section 2
Three Scale-Up Failures — and Why They Happened
|
Failure Mode #1 · Texture / Gelling |
|
“The jam that wouldn’t set”
A fruit preserve with a consistent firm set at 50kg. First 500kg production run: runny or solid depending on where in the vessel it was sampled.
Root cause: Pectin activation is time-temperature sensitive. At production scale, the slow heat-up meant pectin was hydrating differently before reaching set temperature — over-hydrated near the jacket wall, under-hydrated near the centre.
Fix: Validate total time-at-temperature at production scale. Map the thermal profile across the vessel. Adjust cook schedule to replicate the small-scale thermal curve — not the small-scale recipe. |
|
|
Failure Mode #2 · Emulsion Stability |
|
“The dressing that split”
A clean-label salad dressing, stable for 6 months at 20L. First 1,000L trial: separation within 48 hours in the first 200 units filled.
Root cause: The production high-shear mixer generated significantly higher tip speeds than the lab homogeniser. Excessive shear destroyed the xanthan polymer network acting as stabiliser. More shear was applied at larger scale — the opposite of what the team expected.
Fix: Characterise shear input at development scale. Brief the co-manufacturer on the target shear window. Measure droplet size (D50, D90) and viscosity on pilot and production batches. |
|
|
Failure Mode #3 · Food Safety |
|
“The HACCP plan that didn’t transfer”
CCP critical limits validated on a pilot vessel with 8-minute come-up time. Production vessel: 22-minute come-up. Product appearing compliant at the thermocouple was not reaching target temperature at the geometric centre.
Critical: A HACCP plan validated at pilot scale is a working hypothesis, not a validated food safety plan. Every scale transition requires formal revalidation of every CCP against the new process parameters. |
|
|
|
|
Section 3
The Engineering Toolkit for Scale-Up
Successful scale-up isn’t about running more trials until something works. It’s about understanding which physical mechanisms made the product work at small scale, then deliberately recreating those mechanisms at a larger one.
|
|
Dimensionless Analysis — Your Best Prediction Tool
Three numbers cover most mixing and heat transfer scale-up decisions in food manufacturing:
Parameter |
What it tells you |
Scale-up rule |
Re = ρND²/μ |
Laminar vs. turbulent flow regime in the vessel |
Match Re across scales |
P/V = const. |
Mixing energy input per litre of product |
Keep constant (0.1–2 kW/m³) |
Nu = hD/k |
Jacket heat transfer efficiency vs. geometry |
Predict come-up time at scale |
Most food products sit in transitional flow at production scale (Re 10–10,000). Below Re 10 = laminar. Above Re 10,000 = fully turbulent.
|
|
Pilot Trial Data — What Most Companies Miss
Measure this during every pilot trial:
✓ | Temperature at the geometric centre of the product mass (not just jacket outlet) |
|
✓ | Come-up time, hold time, and cool-down time — with timestamps |
|
✓ | Impeller speed (RPM), motor current draw, and approximate tip speed |
|
✓ | Product viscosity at process temperature |
|
✓ | Batch yield and moisture loss |
|
|
|
Briefing a Co-Manufacturer — The Right Questions
→ | Vessel working volume and jacket area (to calculate SA:V ratio) |
|
→ | Agitator type, impeller diameter, and maximum tip speed |
|
→ | Steam jacket maximum pressure and controllable temperature range |
|
→ | Typical come-up time for viscous products in that vessel |
|
→ | CIP system type and validated chemical concentrations and temperatures |
|
|
|
|
|
Section 4
The Quality Trap: When Your Spec Is Impossible at Scale
A product spec built from bench-scale samples applied directly to production batches is structurally unachievable. At small scale, natural variation is tight. At production scale, longer hold times, imperfect mixing, and temperature gradients widen that variation. Your tolerance may be unachievable in 30% of batches — not because the product is wrong, but because the tolerance was never designed for production reality.
|
|
The Fix
Build process capability data at production scale before locking final specifications. Run a minimum of 10–15 production batches and set limits at ±3 standard deviations of the production process — not the bench process.
This is process capability analysis (Cpk). Standard in pharma and industrial manufacturing. Almost never done in food SME scale-up — and one of the most common causes of sustained non-conformance after launch.
|
|
|
|
The Real Test
Are You Actually Ready to Scale?
A business is ready for scale when it can answer four questions clearly — not optimistically. With data, procedures, and a team that can execute without constant intervention.
|
| |
Can we produce consistently?
Is the process documented so a trained operator — not the founder — can run a batch to spec? Do you have process parameters, not just a recipe? Is yield predictable within a defined tolerance?
|
|
| |
Can we package safely?
Has packaging been validated for shelf life and transit conditions — not just the farmers market trestle table? Has migration been assessed for your food contact materials?
|
|
| |
Can we control quality?
Do you have in-process checks that catch problems before the filling line? Are specs based on production-scale data? Is there a defined corrective action when a batch falls out of range?
|
|
| |
Can we fund growth without breaking cash flow?
Larger runs require larger ingredient orders. Retail payment terms are longer than market transactions. Do the unit economics work at volume — and does the cash timing work without a crisis?
|
If the answer to any of these is no, you are still in the engineering phase — and that is the right diagnosis.
|
|
Final Thought
The jump from farmers market to 50,000 units is not a bigger version of the same journey. It is a different game entirely.
At that level, the winners are not just creative brands. They are disciplined operators. They build products that can be repeated, processes that can be trusted, and systems that can carry volume without losing control.
That is the real engineering of scale-up.
|
|
From the Engineer
I’ve seen this pattern play out more times than I’d like. It almost always follows the same sequence: the product is developed with the recipe as the primary document, not a process specification. The engineering parameters — vessel geometry, thermal profile, shear history — are recorded nowhere, or captured informally in someone’s notebook.
Then when it reaches a production facility, those parameters are different — and no one knows by how much, because they were never formally defined. The product “works” on a small scale because the developer is compensating for the process gaps without realising it. Volume removes that compensation.
The farmers market to 50,000 units journey is genuinely exciting. But the gap between those two worlds is mostly an engineering gap — and engineering gaps are solvable, once you name them.
|
|
Process Notes
Practical food engineering, every week.
Have a question, a scale-up story, or a topic you’d like covered?
Reply to this email — it goes straight to me.
processnotes.beehiiv.com
|
ISSUE #004
processnotes.beehiiv.com
|
|
